The technique of pulse-width modulating one or more switched voltages or currents to produce an integrated analog representation of the modulation source has been in use for at least two decades. When endeavoring to achieve high accuracy with this approach, however, a parasitic effect of arises in conjunction with most switching devices. In particular, in addition to the desired switched voltage or current, most switching devices inject a portion of the control signal used to effect the state change.
This lack of isolation manifests as an offset in the output which is not directly indicated by the modulation source. Although this is not usually problematic in unipolar designs such as power supplies, bipolar designs, such as pulse-width-modulated amplifiers, may suffer considerable distortion due to the effect. A need exists to nullify this undesired switching device deficiency.